Predicting the Printability of Cementitious Composites in 3D Printing (2026-03)¶

Printability of 3D printed cementitious materials depends on the pumpability, extrudability and buildability of fresh cementitious pastes, where static yield stress (SYS), dynamic yield stress (DYS) and thixotropy are the most commonly characterized parameters. However, researchers hold convergent opinions in DYS testing while no widely accepted protocols for SYS and thixotropy exist. In this study, a specific set of white Portland cement mixtures incorporating various silica fume (SF) contents were used to prepare cementitious pastes. Three different protocols were employed to measure SYS and thixotropy of 3D printed white cementitious materials (3DPWCMs), aiming to identify their relationships with the structural deformation rate of printed components. Specifically, Kolmogorov-Arnold Network (KAN) model was applied to explore the relevancy between rheological parameters and structural deformation rate, achieving higher predictive accuracy than simple linear regression. Furthermore, published datasets from other cementitious composites were also collected to validate the outputs from the KAN model. Notably, within the tested systems, creep-based protocols appeared to be the most reliable indicators in predicting the structural deformation rate, outperforming both the small amplitude oscillatory shearing and shear-rate-controlled protocols for SYS, as well as the thixotropic loop area and viscosity recovery degree for thixotropy. Our work provides insights into the suitability of various rheological protocols for 3D printed cementitious materials and highlights the potential use of the KAN model in guiding the design and quality control of printable cementitious materials.